* Tags:: #🗞️Articles , [[Modern Data Stack]], [[Data team vision and mission]] * Author:: [[Benn Stancil]] * Link:: [The Modern Data Experience - by Benn Stancil](https://benn.substack.com/p/the-modern-data-experience?utm_campaign=post&s=09&utm_source=pocket-app&utm_medium=share) * Source date:: [[2021-08-20]] * Finished date:: [[2021-08-22]] There is a very formal definition and a different more open one the community is converging to: >Last week, Analytics Twitter melted down over The [[Data Mesh]] ™ and what, exactly, it is. As best I can tell, there are now two definitions. The [original definition](https://martinfowler.com/articles/data-monolith-to-mesh.html) is a fairly complex set of architectural requirements. Its pillars are technical, and, as data mesh creator [Zhamak Dehghani made clear](https://twitter.com/zhamakd/status/1426042889474166792), quite specific. The second definition—the [descriptivist](https://en.wikipedia.org/wiki/Linguistic_description) one the [community](https://twitter.com/juansequeda/status/1427033307984760842) [is](https://cnr.sh/essays/what-the-heck-data-mesh) [gravitating](https://www.linkedin.com/pulse/technology-problem-its-operations-jillian-corkin/?trackingId=H3NVSeacpscJ51FQZbxy7g%3D%3D) [towards](https://twitter.com/petehanssens/status/1427245102334767111)—is more vague, essentially claiming that any decentralized architecture in which teams are responsible for their own “data products” is a data mesh. Another cry, in the line of [[📖 Data Management at Scale]], that our tools are hugely disjointed and we need some centralization layer. The big ones are already doing it: > At Uber, employees can search for a metric, visualize it across different dimensions, and move from code-free exploration directly into writing queries, all while an AI ensures that work doesn’t get repeated. Airbnb built a similar platform, connecting a [[data catalog]] and a metrics repository with a data exploration tool and a SQL IDE. And Netflix designed an entire workflow for creating, sharing, deploying, scheduling, and discovering notebook applications that support everything from dashboards to production models Of course, we don't even know what to trust. This sentence became part of the vision of my team at [[Mercadona Tech]]: > Our goal should be to spend more time debating what to do because of a number on a dashboard than we spend verifying if that number’s right ^f0160e We need to respect the tools people use: > Most notably, [[Excel]] isn’t going away. A modern data experience has to negotiate with it, not treat it as an outdated pariah. And collaboration between roles: > Just as dbt broke down the first wall, a modern data experience needs to break down the others by encouraging collaboration and conversation between business, data, and engineering teams. In other words, in the shorthand of the moment, the modern data experience is purple